About the Project
There are massive challenges with ensuring the quality of temperature-sensitive biological materials, such as blood components, organs, vaccines, and clinical specimens. This challenge exists in clinical laboratory environments and within logistical operations. Even brief temperature fluctuations can result in degradation of samples, financial loss, compromised patient outcomes, and breaches of regulatory compliance. This research proposes the development of a physics-informed machine learning model capable of forecasting cold-chain temperature deviations.
In contrast to conventional monitoring systems that merely record temperature retrospectively, the proposed approach integrates sensor data with thermodynamic modelling to construct a digital twin of temperature-sensitive refrigeration systems. The research will examine how factors such as heat-transfer dynamics, equipment ageing, ambient environmental variability, door-opening behaviour, compressor performance, and storage-load fluctuations shape internal thermal conditions. Machine learning models will be trained to identify early indicators of instability and to predict both short-term and long-term temperature deviations with quantifiable uncertainty. Crucially, the research incorporates an explainable AI (XAI) component, aimed at generating transparent and interpretable model outputs that can be scrutinised by engineers, quality-assurance staff, and regulators.
Techniques such as feature-attribution analysis and physics-guided interpretability will be explored to ensure that predictive insights are trustworthy and auditable. Further work will address robust sensor fusion, anomaly detection, and model interpretability to satisfy regulatory expectations in clinical settings. Empirical validation will be undertaken using controlled test environments alongside real-world datasets from high-risk storage environments. By shifting cold-chain management from reactive, alarm-driven responses towards proactive, predictive, and explainable refrigeration control, the project aims to enhance system reliability, reduce risks, and strengthen the safeguarding of critical biological materials.
Work on this project falls within the larger framework of the NWCAM2 research programme a project supported by the PEACEPLUS Programme, managed by
the Special EU Programmes Body (SEUPB), which will address crucial challenges in the life and health sciences sector by supporting SMEs in the PEACEPLUS region in the development of environmentally sustainable manufacturing processes and products – enabling them to innovate, reduce emissions and compete on a global scale. The PhD will combine statistical modelling, AI/ML, simulation, digital twin and systems thinking to improve cold chain reliability, patient safety, and supply chain resilience for pharmaceuticals, biologics, vaccines, and advanced therapies.
The Role:
The successful candidate will be expected to collaborate with other NWCAM2 partners including University of Ulster, IMR and Trinity College. The broader NWCAM programme will combine deep research with practical implementation to ensure tangible benefits, directly supporting SMEs on both sides of the border to embed advanced technologies such as digital twin, Artificial Intelligence, Machine Learning (ML), and digital modelling. Key responsibilities of the researcher will include:
- Conducting original research on predictive modelling, digital twinning, and cold chain risk forecasting.
- Developing digital twin models representing temperature‑controlled pharmaceutical supply chains.
- Designing and implementing AI/ML algorithms for time‑series prediction, anomaly detection, and risk estimation.
- Building and analysing large-scale datasets, including sensor data, environmental data, and logistics metadata.
- Performing simulation, statistical modelling, and uncertainty quantification to evaluate risk and system behaviour.
- Integrating physics‑based, statistical, and machine‑learning approaches into hybrid modelling frameworks.
- Creating reproducible code, data pipelines, and model documentation that adhere to scientific best practices.
- Conducting model evaluation, including benchmarking, sensitivity analysis, and robustness testing.
- Ensuring compliance with data governance, ethical AI, and regulatory considerations relevant to health‑critical systems.
Essential Qualifications & Skills:
- A minimum 2.1 Honours Degree (or equivalent) in Computer Science, Electronic Engineering, Computer Engineering
- Strong programming ability in Python
- Experience with in NumPy, Pandas, SciPy, scikit‑learn and/or PyTorch/TensorFlow
- Experience handling time‑series data and large, noisy, real‑world datasets
- Hands on experience in Machine Learning
- Solid grounding in probability and statistics
- Strong analytical and problem-solving skills.
- Excellent communication skills
- Ability to write technical reports and prepare scientific publications
- Deliver presentations to diverse audiences
- Work both independently and collaboratively in a research team
- Manage time and research tasks effectively
- Highly self-motivated with a clear interest in interdisciplinary and applied research
- Ability to start position in Ireland immediately and no later than 31st May 2026
- Fluency in English is essential. Candidates whose first language is not English must meet ATU’s minimum English language requirements (e.g. IELTS 6.0 overall, with no component below 5.5 or equivalent)
Contact:
Please send a CV, transcripts, and a cover letter outlining your suitability to the role to: Fiona.Barrett@atu.ie referencing project number and NWCAM2 in subject of email.
For infomal enquiries contact Dr. Kevin Meehan: kevin.meehan@atu.ie
For More Information:
Visit the ATU PhD Page: https://www.atu.ie/research/postgraduate-research-vacancies#temperature-risk-forecasting
